Crystal Generation using the Fully Differentiable Pipeline and Latent Space Optimization
Preprint
Abstract
We present a materials generation framework that couples a symmetry-conditioned variational autoencoder (CVAE) with a differentiable SO(3) power spectrum objective to steer candidates toward a specified local environment under the crystallographic constraints. In particular, we implement a fully differentiable pipeline to enable batch-wise optimization on both direct and latent crystallographic representations. Using the GPU acceleration, this implementation achieves about fivefold speed compared to our previous CPU workflow, while yielding comparable outcomes. In addition, we introduce the optimization strategy that alternatively performs optimization on the direct and latent crystal representations. This dual-level relaxation approach can effectively escape local minima defined by different objective gradients, thus increasing the success rate of generating complex structures satisfying the target local environments. This framework can be extended to systems consisting of multi-components and multi-environments, providing a scalable route to generate material structures with the target local environment.Keywords
All Related Versions
This publication has 26 references indexed in Scilit:
- Space group informed transformer for crystalline materials generationScience Bulletin, 2025
- WyCryst: Wyckoff inorganic crystal generator frameworkMatter, 2024
- Artificial Intelligence Driving Materials Discovery? Perspective on the Article: Scaling Deep Learning for Materials DiscoveryChemistry of Materials, 2024
- Crystal Structure Prediction for Benzene Using Basin-Hopping Global OptimizationThe Journal of Physical Chemistry A, 2021
- Generative Adversarial Networks for Crystal Structure PredictionACS Central Science, 2020
- Homo Citansand Carbon Allotropes: For an Ethics of CitationAngewandte Chemie International Edition, 2016
- New developments in evolutionary structure prediction algorithm USPEXComputer Physics Communications, 2013
- Python Materials Genomics (pymatgen): A robust, open-source python library for materials analysisComputational Materials Science, 2013
- CALYPSO: A method for crystal structure predictionComputer Physics Communications, 2012
- XtalOpt: An open-source evolutionary algorithm for crystal structure predictionComputer Physics Communications, 2011